Instruction Following on ALFWorld Unseen (val)
79.19Success Rate (SR)TCOD (F2B)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TCOD (F2B)Student Model=Qwen2.5-3B2026.04 | 79.19 | 12.47 | |
| TCOD (B2F)Student Model=Qwen2.5-7B2026.04 | 77.61 | 13.16 | |
| Qwen2.5-7B-RL (Teacher)Model Type=Teacher2026.04 | 76.87 | 13.06 | |
| TCOD (F2B)Student Model=Qwen2.5-7B2026.04 | 76.12 | 13.22 | |
| Vanilla OPDStudent Model=Qwen2.5-7B2026.04 | 72.14 | 13.37 | |
| TCOD (B2F)Student Model=Qwen2.5-3B2026.04 | 70.9 | 14.56 | |
| Vanilla OPDStudent Model=Qwen2.5-3B2026.04 | 60.45 | 16.21 | |
| SFTStudent Model=Qwen2.5-7B2026.04 | 48.73 | 20.11 | |
| SFTStudent Model=Qwen2.5-3B2026.04 | 25.37 | 24.16 | |
| Zero-ShotStudent Model=Qwen2.5-7B2026.04 | 8.96 | 28.46 | |
| Zero-ShotStudent Model=Qwen2.5-3B2026.04 | 2.24 | 29.63 |